Merging lines of an array

Question

I've got an array like this, which is a result from a Google analytics request. I asked for the amount of visits for the last three months.

$statPerMonth
array (size=2)
 '08' => // The month (August)
 array (size=34)
 0 => 
 array (size=3)
 0 => string '08' (length=2) // Month again
 1 => string 'admin.testweb.fr' (length=19) // host
 2 => string '1' (length=1) // amount of visits
 1 => 
 array (size=3)
 0 => string '08' (length=2)
 1 => string 'audigie-espace-auto.reseau-fivestar.fr' (length=38)
 2 => string '6' (length=1)
 2 => 
 array (size=3)
 0 => string '08' (length=2)
 1 => string 'www.audigie-espace-auto.reseau-fivestar.fr' (length=31)
 2 => string '9' (length=1)
 3 => 
 array (size=3)
 0 => string '08' (length=2)
 1 => string 'carrosserie-abberis.reseau-fivestar.fr' (length=38)
 2 => string '7' (length=1)
'07' => 
 array (size=47)
 0 => 
 array (size=3)
 0 => string '07' (length=2)
 1 => string 'www.anothersite.testweb.fr' (length=13)
 2 => string '1' (length=1)
 1 => 
 array (size=3)
 0 => string '07' (length=2)
 1 => string 'admin.testweb.fr' (length=16)
 2 => string '2' (length=1)
 2 => 
 array (size=3)
 0 => string '07' (length=2)
 1 => string 'admin.testweb.fr' (length=19)
 2 => string '1' (length=1)
 3 => 
 array (size=3)
 0 => string '07' (length=2)
 1 => string 'audigie-espace-auto.reseau-fivestar.fr' (length=38)
 2 => string '20' (length=2)
 4 => 
 array (size=3)
 0 => string '07' (length=2)
 1 => string 'www.admin.testweb.fr' (length=19)
 2 => string '1' (length=1)

This array represents the amount of visits for my websites but you can see that the values [‘08’][‘1’] and [‘08’][‘2’] are identical (only 'www.' differs) I want to merge those cells and add their value (because it’s the same site !) in order to get the total amount of visits for a site with it’s two hostnames.

Consider $sites as an array of Site Object (websites). the getHost() method will return the site host for example ‘my-host.fr’ without the 'www' consider $statsPerMonth array explained above

Finally, consider this algorithm:

foreach ($statsPerMonth as $actualMonth => $stats) {
 foreach($sites as $site) {
 $siteHost = $site->getHost();
 foreach ($stats as $row) {
 if (strstr($row['1'], $siteHost)) {
 if(isset($globalStats[$actualMonth][$siteHost])) {
 $globalStats[$actualMonth][$siteHost] = $globalStats[$actualMonth][$siteHost] + $row['2'];
 } else {
 $globalStats[$actualMonth][$siteHost] = 0;
 $globalStats[$actualMonth][$siteHost] = $globalStats[$actualMonth][$siteHost] + $row['2'];
 }
 }
 if(!isset($globalStats[$actualMonth][$siteHost])) {
 $globalStats[$actualMonth][$siteHost] = 0;
 }
 }
 }
 }

This algorithm returns the $globalStats array in this form:

array (size=3)
 '08' => 
 array (size=43)
 'carrosserie-la-cascade.reseau-fivestar.fr' => int 1
 'audigie-espace-auto.reseau-fivestar.fr' => int 15
 'carrosserie-abberis-fivestar.fr' => int 16
 'carrosserie-arenales-jonathan.reseau-fivestar.fr' => int 0
'07' => 
 array (size=43)
 'carrosserie-la-cascade.reseau-fivestar.fr' => int 2
 'audigie-espace-auto.reseau-fivestar.fr' => int 20
 'carrosserie-abberis-fivestar.fr' => int 0
 'carrosserie-arenales-jonathan.reseau-fivestar.fr' => int 4
'06' => 
 array (size=43)
 'carrosserie-la-cascade.reseau-fivestar.fr' => int 0
 'audigie-espace-auto.reseau-fivestar.fr' => int 29
 'carrosserie-abberis-fivestar.fr' => int 0
 'carrosserie-arenales-jonathan.reseau-fivestar.fr' => int 4

This is exactly what I want but I think we can improve this algorithm to make it more efficient (because the arrays are big). Any ideas about how to make this algorithm better?

• DFord: Are the sites returned in any specific order?

Answer 1

The innermost loop body can be simplified. Instead of this:

if (strstr($row['1'], $siteHost)) {
 if(isset($globalStats[$actualMonth][$siteHost])) {
 $globalStats[$actualMonth][$siteHost] = $globalStats[$actualMonth][$siteHost] + $row['2'];
 } else {
 $globalStats[$actualMonth][$siteHost] = 0;
 $globalStats[$actualMonth][$siteHost] = $globalStats[$actualMonth][$siteHost] + $row['2'];
 }
}
if(!isset($globalStats[$actualMonth][$siteHost])) {
 $globalStats[$actualMonth][$siteHost] = 0;
}

I think this is equivalent:

if (strstr($row[1], $siteHost)) {
 if (!isset($globalStats[$actualMonth][$siteHost])) {
 $globalStats[$actualMonth][$siteHost] = 0;
 }
 $globalStats[$actualMonth][$siteHost] += $row[2];
}

(I also removed the quotes around the array indexes 1 and 2, I don't think you needed them.)

But the bigger improvement will be to cur out the foreach($sites as $site) loop. Think about it, for each site on your list, you re-process all the arrays for the given month. It would be better to process the stats only once. For this you will need a helper function getCanonicalSiteName that will give you, for example, audigie-espace-auto.reseau-fivestar.fr for both audigie-espace-auto.reseau-fivestar.fr and www.audigie-espace-auto.reseau-fivestar.fr in the processed array. Something like this (untested):

foreach ($statsPerMonth as $actualMonth => $stats) {
 foreach ($stats as $row) {
 $siteHost = getCanonicalSiteName($row[1]);
 if (!isset($globalStats[$actualMonth][$siteHost])) {
 $globalStats[$actualMonth][$siteHost] = 0;
 }
 $globalStats[$actualMonth][$siteHost] += $row[2];
 }
}